set fit errorvariables

f(x)=a*exp(-(x-x0)/c)
a=1000;
x0=328;
c=10;
fit [328:] f(x) 'out.dat' u 1:2 via a,c
print c,c_err